How Axyo works, and where it applies.

Mid-market manufacturers and distributors that carry inventory and run multiple sites, where planning still depends on spreadsheets or on an ERP module that was never built for it. The people in it every day are demand, supply and inventory planners and buyers; the people who care about the result are the operations and finance leaders who own service, margin and working capital.

No. Axyo sits above ERP and systems of record. It can replace spreadsheet-heavy or legacy planning workflows for demand, supply, inventory, financial planning and supplier performance decisions, then writes approved actions back through controlled paths.

It forecasts, and it has dashboards, but neither is the point. A report tells you what happened. Axyo tells you what is about to go wrong, what it will cost, how long you have to act, and what to do — with the evidence attached and a person making the call. Most customers keep their BI stack alongside it for enterprise reporting.

Traditional IBP aligns demand, supply and finance on a meeting cycle. Between those meetings, nothing is watching. Axyo runs continuously — sensing internal change and external signals as they arrive, at the grain of 130+ named loss scenarios rather than at plan level — so an exposure surfaces as a quantified early warning while there is still time to prevent the loss.

Axyo is a decision layer, not a transaction system. It does not run purchase-to-pay, sourcing events or auctions, supplier discovery, or invoice processing. Pricing and treasury decisions stay with those teams — Axyo gives them the evidence rather than making the call. Approved actions are written back to your systems of record, and the transaction still executes there.

ERP, MRP/MES, WMS/TMS, CRM, finance and BI, data warehouses and spreadsheets, plus external signal feeds. Read access is enough to begin, and write-back is optional and approval-controlled. Which connectors apply to you, and how, is confirmed against your systems and security constraints during the technical review rather than assumed up front.

Six fields carry the joins that everything else hangs off: item SKU, on-hand quantity, purchase order number, ordered quantity, site code, and supplier ID. Add demand or shipment history and the forecast comes with it. More data sharpens the answer; those six are what make it possible at all.

It is expected, not disqualifying. Axyo profiles what arrives, scores each field for completeness and freshness, quarantines rows that fail, and tells you what is usable before anything is modeled. No platform can produce a trustworthy recommendation from unusable data, so we would rather show you the gap than model around it.

Yes, and most teams do. Start with the planning spreadsheets your planners already work in, plus straightforward exports — item and supplier master, an on-hand inventory extract, open purchase orders, and demand history. CSV or Excel is enough; nothing has to be installed. You validate the output on your own numbers first, then connect ERP and other source systems once it has earned it.

Yes. Supplier events, freight and port activity, weather, commodity cost, FX, tariffs, macro indicators and market signals, where they improve forecast, risk or recommendation quality. Each signal is tied to the specific items, suppliers, sites, lanes and customers it affects.

Time-series forecasting, regression and classification through AutoML. The 30+ algorithm library refers specifically to time-series forecasting: Axyo profiles each history for trend, seasonality, intermittency, volatility and structural breaks, then selects the model that fits that item and horizon.

For explanation, summarization, drafting and guided analysis. Deterministic calculations, thresholds, approvals and write-back logic remain governed and reproducible. A generative model explains a number; it is never where the number comes from.

Backtesting and holdout validation on your own history before anything goes live, then accuracy tracked in production by item, location, customer and horizon. Accuracy is reported by domain and horizon rather than as one aggregate figure, because a single number hides where a model is weak.

The number says so. Below a coverage threshold Axyo shows a measure as not rated rather than as a confident figure, and a supplier score without enough evidence reads Insufficient Data. Thin evidence is disclosed on the number itself, not in a footnote.

No. Forecasting models are fit on your own history, item by item, and stay within your tenant. Your data is isolated at the tenant level and is not pooled with anyone else’s or used to improve another customer’s results.

Yes, and they should. Every recommendation can be accepted, changed, rejected or escalated. It is a draft until a person commits it.

A signal has to be corroborated before a case opens at all — the working rule is two signals plus one corroboration. Materiality then decides which of those are worth anyone’s attention, and runway separates time to impact from time left to act. The output is the few cases that matter this week, not a feed.

Shared SaaS is the default. VPC deployment is available for customers with stricter isolation needs. On-premise can be considered by exception, after infrastructure sizing and a support model review.

Deployment and data-residency patterns can be aligned to locale and compliance needs, including US, India and APAC requirements. The specifics are confirmed as part of the technical review rather than assumed.

Axyo is SOC 2 Type II certified — an independent audit of how the controls actually operated over a period, not a self-assessment — and the controls and trust documentation are available for your review. Day to day that means single sign-on, role-based permissions with least-privilege access by user and workflow, tenant-level data isolation, environment separation, and privacy-aware data handling by role and process.

Because Axyo recommends decisions rather than only reporting on them, the audit trail runs into the product itself. Every number carries its source lineage back to the record it came from, and who saw, changed, approved or wrote back a recommendation is retained for review. Approvals, thresholds and write-back logic stay governed and reproducible.

It depends on your data, your systems and the scope, so we do not quote a date before seeing them. The sequence is consistent: discovery and data assessment, a file-based pilot on your own data, then integration and controlled write-back.

A loss assessment. You share twelve months of history, read-only. We come back with your exposure, line by line, on your own numbers — no installation, no commitment.

Axyo is sold as an annual subscription. We confirm the commercial shape with you once we understand your systems, your data volume and which decisions you want covered — we would rather scope it properly than quote a number that has to change later.

That is the intended path. A focused pilot on your data, on one decision that is painful today, before anything is connected to a production system.